calendar | intranet

SIMVIR project

Description : The SIMVIR research and development project, gathering in a consortium 7 partners, and planned from 2012 to 2014, focuses on the creation of 3D geo-specific digital models for simulator studies and driver training.

Partners : Oktal, Virtuel City, ISM/CRVM, PSA, Keolis, Pays de Montbéliard, Armines

Years : 2012-2014

Web site : SIMVIR

Major industrial companies have been using driving simulation for a long time both upstream in the development of vehicles and downstream for driver training. Thus, automobile manufacturers are using driving simulators allowing them to reproduce at will and without risk vehicle tests in both phases of innovation and development. Similarly, major rail operators and more recently Tramway operators, implement simulators for initial and in-service training of their drivers.

In both cases, these "heavy simulations" require a perfectly faithful reproduction of the road and rail environment with very high standards in visual and functional representation of the environment.

On the other hand and more recently, 3D Learning is a new form of teaching that uses a virtual representation of the real environment as a medium of communication. Its components include various forms of presentation, management of the learning experience, and a virtual learning community of content developers and experts. 3D Learning provides faster learning at a lower cost and without security risks, it facilitates access to learning and allows detailed monitoring of all participants at each stage of the learning process.

At the rate the culture and knowledge evolve, organizations increasingly use 3D Learning that allows their employees to acquire a real technical skill and a thorough knowledge of the physical processes to perform: this provides to these companies and organizations a major advantage in terms of safety and quality. Technologies implemented in the 3D Learning now tend to come close to "heavy simulation" but without reaching the high level of expectation.

However, in both cases, the technological limitations of automated databases representing a real environment, the difficulties due to detailed and faithful representation on a large scale, the scanned data and the limitations of display technologies inhibit the growth of simulator and 3D Learning.

Obviously, some recent research works (especially the line of TerraNumerica projects) have clearly shown and open the way for new methods to further automated production without however allowing a fully automatic process. These new methods are based on the acquisition and integration of geo-referenced data. On urban buildings, acquisition methods disseminated by the project (from the ground and air) have already been industrialized by VirtuelCity. At the same time, OKTAL which for more than 20 years, has continued to develop and enrich its own chain of road and rail production environment, incorporated in 2005 with PCA Peugeot Citroën a semi-automated production from geo-referenced data sources to driving simulators.

With the support of two key users (PCA Peugeot Citroën and Keolis) OKTAL clearly aims to enrich virtual environment production tools for driving simulators:

  • By integrating both demonstrated viable solutions through TerraNumerica projects: for example, automated acquisition of building and road surface via VirtuelCity,
  • By extending and specializing the automatic acquisition methods for road and railway signaling with ARMINES
  • Adapting and integrating civil modeling tools,
  • By integrating parameters based on perceptual criteria (with ISM) applications and functional criteria (with end users).

The objective for users involved in the project is to drastically reduce the cost of production of virtual environments while improving the quality of modeling. This is the "key" challenge of an industrial deployment solution for OKTAL and VirtuelCity.

Automotive or Railway simulations, particularly in urban environments, require a wealth of characters: artificial intelligence for both traffic and pedestrians / passengers. For a long time, driving simulators incorporate behavioral models that have been the subject of much research. The OKTAl SCANeR™ solution is the result of over 15 years of research and development and has benefited many technology transfer, involving both Renault and Ifsttar. The product is now recognized worldwide as being very advanced in modeling traffic and pedestrians in real time driving applications, as evidenced by numerous international publications.

In the continuity of TerraNumerica projects, the project "Training Bus" explored the possibility of producing a "serious game" exploiting these virtual environments. At the same time and on the same "serious gaming" call for proposals OKTAL demonstrated through the "geDriver" project that a serious game (eco-driving in this case) could be produced and distributed using technology professional simulation (SCANeR ™ software in this case).

With these references, the "SIMVIR" project is not intended to develop new models of Artificial Intelligence, but to use force and existing models by expanding the number of players in a driving session. We know that these constraints impose to enrich the environment RoadXML numerical model and to implement original scripting methods.

Thus, and in the same way, the goal for the users involved in the project is to reduce the time required for the development of driving scenario including a large number of actors (traffic, pedestrians and passengers).

Finally, the automated integration of realistic data at a pedestrian level exceeds the strict interest for automotive and tramway simulators. Industrializing these integrated data and optimizing the display at a high level of detail are fantastic opportunities to enhance the architectural heritage of French cities.

In this respect, the consortium intends to use its "professional" research program to enhance its work with the possibility of large public data dissemination on IPad / Android Tablets, optimized for industrial simulators. The territory of Montbéliard (Urban Community of 29 municipalities), location of VirtuelCity and PCA Peugeot Citroën, will serve as a prototype. This mixed territory, urban, semi-urban and rural is the expression of a rich experience to test driving simulators, public transport and to promote French heritage to the public.

Technologically, dissemination will be done in the territory on Tablets/IPads through a prototype available to cultural services and general public.

In continuation of recent projects to digitize the territory and on the basis of results of research projects on driving simulation crystallized by OKTAL on SCANeR ™ and OKSimRail ™ products, "SIMVIR" research and development project aims to integrate and optimize tools to produce a large-scale (several tens of urban kilometers) 3D digital model with an accuracy of one centimeter to be faithful to reality, adapted to the needs of users operating simulators and/or working in the 3D Learning.